Anticipatory conditioning of spelling-to-sound translation

College students’ pronunciations of initial c and g were examined in English words and nonwords, both monosyllables and polysyllables. Pronunciations were influenced by adjacent context—whether the following letter was e or i—and by long-distance context—whether the item contained a suffix or spelling pattern characteristic of Latinate words. Pronunciations were also influenced by whether students had studied a language such as French in which c and g are systematically fronted before those vowels. The findings were not well fit by either a dual-route or a single-route reading model. Although skilled readers were affected by the contextual patterns in the input, they did not use these patterns to the extent that would be expected given the patterns’ reliability in the English vocabulary. The results are discussed in terms of the size of the window within which spelling-to-sound translation takes place and the nature of the units that are used.